[00:04] <kim0> weird networking troubles: http://ubuntuforums.org/showthread.php?t=1595384
[00:08] <ablert> smoser, thanks! 10.04 (lucid) is what i was seeking
[00:26] <smoser> ablert, so, when that bug is fixed, you'll be able to get it in backports
[00:26] <smoser> and, in the future, that is the eventual landing place for such things
=== ank_ is now known as ank
[01:44] <maruq> hi guys
[01:44] <maruq> I'm getting high load issues on 10.04, running on EC2
[01:44] <maruq> I get the lovely " System information disabled due to load higher than 1" message
[01:45] <maruq> and it shows up as load of ~1.20 in Ganglia
[01:45] <maruq> but nothing really running in top
[01:56] <flaccid> maruq: known bug in kernel
[01:56] <flaccid> see https://bugs.launchpad.net/bugs/574910
[01:56] <uvirtbot> Launchpad bug 574910 in linux-ec2 "High load averages on Lucid while idling" [Undecided,In progress]
[01:56] <maruq> ah okay, thanks
[01:57] <maruq> I Googled & saw a few things, but no real answer
[01:57] <maruq> most people seemed to restart & fix, but doesn't work in my case
[01:57] <flaccid> yeah its an accounting bug
[01:58] <maruq> so no fixes at the moment?
[01:59] <flaccid> see end of ticket, there are kernels published
[02:14] <maruq> flaccid: thanks, just reading through things now ;p
[02:16] <flaccid> coolio
[02:36] <maruq> flaccid: there's a lot of random in that thread, but made it through
[02:37] <flaccid> pretty typical of these kind of bugs :)
[02:37] <maruq> so using one of the test kernels should fix my load issue for now?
[02:37] <maruq> us-east-1 aki-84b75ded
[02:37] <flaccid> yep
[02:38] <maruq> and then just update to the latest AMI when it comes out - hopefully a couple of weeks, judging by date of post
[02:38] <maruq> which should hopefully fix the t1.micro issue I was having as well ;p
[02:38] <maruq> (Java = death)
[02:42] <flaccid> hopefully maverick will fix all this, and there is daily maverick you can test already
[02:43] <maruq> yeah I tested with Maverick, but unfortunately chef server doesn't install on it (rabbitmq config issue)
[02:43] <maruq> so I guess I wait until the Chef guys fix their boot stuff, then jump to Maverick ;p
[02:44] <flaccid> or just use debian squeeze :p
[02:54] <maruq> okay, booted a node with that kernel & now it's running all the chef stuff
[02:54] <maruq> I'll give it a little while to idle & let you know how things go
[02:56] <maruq> thanks
[02:58] <flaccid> cool np
[06:15] <maruq> flaccid: just checked the top on that kernel, it's still looking high
[06:16] <maruq> top - 05:16:08 up  3:25,  1 user,  load average: 1.10, 1.05, 1.00
[06:16] <maruq> Tasks:  61 total,   2 running,  59 sleeping,   0 stopped,   0 zombie
[06:17] <flaccid> what percent cpu?
[06:18] <maruq> mountall seems to be running at 42%, everything else is 0%
[06:19] <flaccid> that could by one. you'd wanna check top under root
[06:19] <flaccid> once you have it idle, see if you still get the high load
[06:22] <maruq> what would be the best way to get it to idle... I'm assuming mountall is probably fairly important process, so best not to kill?
[06:23] <flaccid> question is why is mountall running and what called it?
[06:25] <maruq> sorry, what's the best way to find that out?
[06:25] <flaccid> could be in an rc script
[06:25] <maruq> it's being ran by root
[06:26] <flaccid> you might want to check that the fstab entries are correct
[06:26] <flaccid> and work out which is having the mount problem
[06:26] <maruq> an ps aux shows it's ran as "mountall --daemon"
[06:26] <flaccid> oh right
[06:26] <flaccid> maybe ubuntu does this thing as a daemon
[06:26] <flaccid> see man mountall
[06:27] <flaccid> and see if there is such a service
[06:31] <maruq> there seems to be stuff in /etc/init
[06:31] <maruq> mountall.conf, mountall-net.conf, mountall-reboot.conf, mountall-shell.conf
[06:32] <maruq> mountall.conf has "start on startup", "stop on starting rcS"
[06:32] <flaccid> you'll need to research that. its bad that it is taking resources like that
[06:32] <maruq> and "expect daemon"
[06:33] <maruq> runs a script block containing "exec mountall --daemon $force_fsck $fsck_fix"
[06:33] <flaccid> i don't agree with implementing such a daemon, but this is what ubuntu obviously has done
[06:35] <maruq> from Googling, it looks as though erichammond may have similar thing
[06:35] <maruq> http://alestic.com/2010/09/ec2-bug-mountall
[06:37] <maruq> testing if that's the case now
[06:38] <maruq> yep, that seems to have fixed it
[06:38] <maruq> mountall is no longer stealing cpu
[06:44] <maruq> top - 05:44:34 up 7 min,  1 user,  load average: 0.04, 0.31, 0.18
[06:44] <maruq> ^^ looking a bit better there ;p
[06:44] <uvirtbot> maruq: Error: "^" is not a valid command.
[06:45] <maruq> looking a bit better there
[06:45] <maruq> I'll let it play out for the 15mins or so, but looking good so far
[06:46] <maruq> thanks
[06:59] <maruq> top - 05:59:28 up 21 min,  1 user,  load average: 0.00, 0.00, 0.05
[06:59] <maruq> okay, think that's fixed my issues
[06:59] <maruq> thanks flaccid & erichammond
=== mathiaz_ is now known as mathiaz
=== xfaf is now known as zul
[15:37] <daker> kim0, anything about c.u.c ?
[15:38] <kim0> daker: yeah indeed .. we got some more refinements to do
[16:48] <smoser> erichammond, do you know (or anyone)
[16:48] <smoser> ec2-run-instances; ec2-stop-instances ; ec2-start-instances
[16:48] <smoser> i'm guessing each stop-start cycle is a new full hour charge
[17:32] <erichammond> smoser: I am under the same assumption that each stop/start begins a new hour.  I don't remember if I got that from a reliable source or just made it up.  I might test it with a spare, unused account if I have a few minutes.
[17:33] <smoser> testable after lag of account updates
[17:33] <smoser> could even use ap-southeast-1 (or some other region you dont really use for that account)
[17:33] <erichammond> I have any number of accounts that I created for test purposes and don't use at all :)
[17:33] <smoser> i just did several start-stop cycles and thought "hey, this is costing me"
[17:33] <smoser> erichammond, do you use consolidated billing ?
[17:34] <smoser> i've considered doing that for my multiple accounts
[17:34] <erichammond> smoser: No consolidated billing.  I only have a few accounts that get charged reguarly, and one of those is like a penny a month.
[17:36] <smoser> erichammond, one more question
[17:37] <smoser> with IAM, i'm under the impression there is no way to differenciate who launched an instance, right  ?
[17:37] <smoser> ie, i can't really see that one account did 95% of the launches and another did 5 or something
[17:38] <erichammond> smoser: correct
[17:38] <smoser> they've got some work to do on IAM still
[17:39] <erichammond> RightScale has been able to track and report that for years.
[17:39] <erichammond> As long as you don't give people the AWS keys and make them use RightScale accounts to create resources.
[17:39] <erichammond> I imagine it's a commonly requested feature of AWS.
[17:39] <erichammond> since there are so many big companies using it now.
[17:40] <smoser> yeah. but then you're either a.) interfacing with rightscale's api or b.) not interfacing with an api
[17:40] <erichammond> exactly
[17:41] <erichammond> With a few tweaks to rules and triggers, it might be possible to use tags to identify user ownership and only let users control instances they created.
[17:41] <erichammond> Amazon will probably come up with a more comprehensive approach, though it may take another year or two :)
[17:42] <erichammond> (when I say "tweaks" I mean AWS enabling new features, not things we can do today)
[17:42] <smoser> erichammond, are you thinking you'll be able to attend (i'm assuming remotely) any UDS-N ?
[17:42] <smoser> other than you being a grumpy stick in the mud, your voice is appreciated :)
[17:43] <erichammond> I'm not traveling to Florida, but I should be able to attend afternoon sessions given the time zone.
[17:44] <smoser> http://ubuntu-smoser.blogspot.com/2010/09/using-policies-in-aws-identity-and.html outlines my failed attempt at sharing a ec2 account with naming conventions
[17:44] <smoser> i didn't think of doing it based on tags
[17:46] <erichammond> Seems like tags would help reduce mistakes even if they can't yet be used to enforce rules.
[18:04] <erichammond> smoser: stop/start charge test is in progress.  Waiting for results to show up on the activity page.
[18:08] <erichammond> blog post written and waiting for the results :)
[19:20] <timwood> Hi.  Does anyone know how to download an Amazon AMI image from S3 for use/conversion with a local UEC cloud?
[19:23] <timwood> Or if it's even possible?  I'd love to use a public AMI that someone has created.
[19:23] <timwood> in our local cloud
=== Hussain is now known as Hussain_Away
[19:43] <timwood> anyone?
[19:51] <smoser> timwood, its not really possible.
[19:51] <smoser> you can launch the instance, then rebundle it
[19:51] <smoser> and then take your rebundled image
[19:51] <smoser> but you dont have access to the pristine images on ec2 that you do not own
[19:51] <smoser> for ubuntu images, we make pristine images available to the public
[19:51] <smoser> http:/uec-images.ubuntu.com
=== xfaf is now known as zul
[20:02] <timwood> smoser: thanks for the response.  It's a shame, I'd really like to get this particular image.  Maybe I'll contact the developer and see if they are willing to let me download it.
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates